P-P and P-S inversion of 3-C seismic data: Blackfoot, Alberta

Transcription

1 P-P and P-S inversion of Blackfoot 3-C P-P and P-S inversion of 3-C seismic data: Blackfoot, Alberta Robert J. Ferguson ABSTRACT P-P and P-S inversion was applied to the vertical and radial components of the Blackfoot 3-C survey, to yield P velocity, S velocity and Vp/Vs. This work was done as part of an ongoing inversion project which will compare the Velocity estimates from different frequency ranges. The results of the work to this point indicate that for the 10 Hz data, both the P-P and P-S inversion methods provide good velocity values. The resulting Vp/Vs estimates showed a Vp/Vs decrease over the zone of interest. Preinversion processing flows were developed for both the P-P and P-S data. INTRODUCTION Pre-stack seismic data were acquired from the data set, with preliminary processing applied. The data was then processed for inversion using techniques that reduced random noise, positioned the data correctly, and broadened the frequency band. For the P-P data this was a simple process - P-P inversion can done post-stack. The P-S data however, required a more unusual approach. Because the P- S inversion algorithm used in this experiment was applied pre-stack, processing goals had to be met pre-stack as well. PROCESSING OF P-P DATA The P-P data was obtained as a set of CDP gathers with the following processing applied (Gorek, 1995): Geometry Assignment Trace Edits Shot-mode FK Filter Spherical Divergence Surface-Consistent Spiking-Decon TV Spectral Whitening Statics NMO The P-P data were then processed for input to inversion. The goal of this processing was to balance the reflection amplitudes, improve the signal to noise ratio, and position the reflection energy correctly. This was achieved using the following flow: Surface Consistent Amplitude Correction - Shot, Receiver, Channel, CDP CREWES Research Report Volume 7 (1995) 41-1

2 Ferguson CDP Stack FX Decon Phase Shift Migration Spiking Decon Figure 1 is the resulting migrated section. The target Glauconitic Channel (Miller et al, 1995) can be seen to the right of the well, at the issippian level. INVERSION OF P-P DATA Post-stack inversion was performed on the P-P section. P sonic logs from wells and were projected onto the line (Miller et al, 1995), correlated to the P-P section in time (Figure 2) and were used as initial P velocity estimates. Inversion testing was done using 3 different P-P inversion algorithms available in Promax. Inversions compared were: Time Domain Recursive, Frequency Domain recursive and Sparse Spike. The Frequency Domain Recursive inversion was found to provide the most stable results over the zone of interest, and was therefore selected for inversion of the P-P data. The following P-P inversion flow was followed. Band-Pass filter ( ) Frequency Domain Recursive Inversion - Frequency Ranges to Honour: Hz seismic, 0-10 Hz initial estimate Figure 3 shows the resulting P velocity estimate, with wells and overlain for comparison. Note the clearly resolved channel to the right of well ). PROCESSING OF P-S DATA The P-S data was obtained as a set of CCP gathers with the following processing applied (Gorek et al., 1995): Geometry Assignment Asymptotic Binning, Vp/Vs=2.22 Trace Edits Shot-mode FK Filter Spherical Divergence Surface-Consistent Spiking-Decon TV Spectral Whitening Statics 41-2 CREWES Research Report Volume 7 (1995)

3 P-P and P-S inversion of Blackfoot 3-C NMO The P-S data were then processed for input to inversion. The goals of this processing, as for the P-P data, were to balance the reflection amplitudes, improve the signal to noise ratio, and position the reflection energy correctly. Because the P-S inversion is applied pre-stack, processing had to be applied pre-stack. This was accomplished by sorting and then stacking the data into common-offset panels, and applying each processing step to each offset panel. The following processing flow was used: Common Offset Binning: Bin width = 40m FX Decon of Common Offset Planes Phase Shift Migration of Common Offset Planes Discard Offset s: 0m - 400m Spiking Decon of Common Offset Planes Figure 4 is a stack section of the P-S data, offsets m were discarded as they degraded the quality of the stack. A comparison of this section to the equivalent section from Gorek, 1995, shows that pre-stack application of noise reduction, migration and signal enhancement achieves almost identical results to those of post-stack application. INVERSION OF P-S DATA Inversion was performed on the P-S section. A detailed description of the P-S inversion algorithm is described in Ferguson, Input to the P-S inversion is summarized below: CCP gathers P velocity correlated to P-S time Initial S velocity The CCP gathers were obtained using the processing flows described above. The P velocity used was the P velocity of figure 3, and was correlated to P-S time using the horizon interpretations of Figures 5 and 6. During this process, a sinc function interpolator was used to resample the stretched P velocity. Initial S velocity estimates were derived for the two well locations, by scaling the P sonics with the Vp/Vs ratios implied by the horizon picks of Figures 5 and 6. The resulting S sonics were correlated to the P-S data using the same process as for the P inversion velocity - using the horizon picks to stretch the velocity followed by sinc function interpolation. The P-S inversion proceeded as follows: Band-pass Filter: P-S Inversion: Reflection scalar = 0.05, Error Constraint = 15% CREWES Research Report Volume 7 (1995) 41-3

4 Ferguson Figure 7 shows the resulting S velocity estimate, with wells and overlain for comparison. The target channel is much less evident on this section than on the P-P inversion. The velocities of Figures 7 and 3 were then used to compute a Vp/Vs section. Figure 8 is a Vp/Vs section derived from the P-P and P-S inversions. The ratios tend to fall into the range of 1.5 to 2.5, with anomolously high values up to 9.0 below the level. Figure 9 is a close-up of the Vp/Vs section, amplitudes above 3.0 are clipped. This close-up reveals much greater detail through the zone of interest (1400 ms ms) and shows areas of Vp/Vs decrease in the zone just above the issippian. CONCLUSIONS The vertical and radial components of the line were inverted to provide estimates of P velocity, S velocity and Vp/Vs. It was found that P-P and P-S inversion provided good velocity estimation, and the resulting Vp/Vs showed decreases over the zone of interest. Pre-inversion processing flows were developed to be employed in further inversion work. REFERENCES Ferguson, R. J., and Stewart, R. R., 1995, Constrained inversion of P-S seismic data: CREWES Project Research Report, v.8, chpt. 18. Gorek, S. J., Stewart, R. R., and Harrison, M. P., 1995, Blackfoot processing: CREWES Project Research Report, v.8, chpt. 38. Miller, S. L. M., Aydemir, E. O., and Margrave, G. F., 1995, Preliminary interpretation of P-P and P- S seismic data from the Blackfoot broad-band survey: CREWES Project Research Report, v.8, chpt CREWES Research Report Volume 7 (1995)

5 P-P and P-S inversion of Blackfoot 3-C P-P Migrated Stack Vik Wab Fig. 1. Vertical component section. This display has been filtered and AGC scaled for viewing only. No AGC scaling was applied prior to inversion. P sonic log locations of wells and are indicated. The Glauconitic Channel (Miller et al, 1995) can be seen to the right of the well, at the level. CREWES Research Report Volume 7 (1995) 41-5

6 Ferguson P sonics Correlated to P-P Section Viking Wab Fig. 2. Correlation of P sonics to P-P stack section. The resulting stretched P sonics were used as initial velocities for P-P inversion 41-6 CREWES Research Report Volume 7 (1995)

7 P-P and P-S inversion of Blackfoot 3-C P-P Inversion Vik Wab Fig. 3. Inversion of P-P data. P sonics and are overlain for comparison. The CDP range is 101 on the left and 251 on the right. The target channel, to the right of at the level, is clearly resolved. CREWES Research Report Volume 7 (1995) 41-7

8 Ferguson P-S Migtrated Stack Vik Wab Fig. 4. P-S stack section. FX decon, Migration and signal enhancement all done pre-stack CREWES Research Report Volume 7 (1995)

9 P-P and P-S inversion of Blackfoot 3-C Fig. 5. Horizon picks for P-P data. These, in conjunction with the corresponding P-S horizon picks, were used to correlate P velocity and S-velocity estimate to P-S time for P-S inversion. Fig. 6. Horizon picks for P-S data. CREWES Research Report Volume 7 (1995) 41-9

10 Ferguson P-S Inversion Vik Wab Fig. 7. Inversion of P-S data. S sonics and are overlain for comparison. The CDP range is 101 on the left and 251 on the right. The target channel, to the right of at the level, is much less evident than on the P-P inversion CREWES Research Report Volume 7 (1995)

11 P-P and P-S inversion of Blackfoot 3-C Vp/Vs Section Vik Wab Fig. 8. Vp/Vs section. Values generally fall within reasonable levels, , with anomolously high values just below the. CREWES Research Report Volume 7 (1995) 41-11

12 Ferguson Vp/Vs Section Vik Fig. 9. Close up for Figure 8, with Vp/Vs ranges clipped at 3.0. Vp/Vs decrease visible across the target zone CREWES Research Report Volume 7 (1995)